Sampling/ Design Effect and ICC
Cross-Cutting Concept

Design Effect and ICC Explained

What the design effect and intraclass correlation actually mean, and how to set them correctly for cluster and multi-stage sampling.

6 min read DEFF and ICC Intermediate

People who live in the same village tend to be more alike than two people picked at random from the whole population, similar access to services, similar local conditions, similar exposure to the same programmes. That similarity has a direct, calculable cost in required sample size. Design effect and ICC are how that cost is calculated.

01Why clustering costs sample

Simple random sampling assumes every observation provides fully independent information. Cluster sampling does not. Once something is known about one person in a village, a little is already known about their neighbour, simply because they share a context. Each additional person from the same cluster adds less new information than a fully independent person would.

The design effect quantifies exactly how much less. A DEFF of 1.8 means 80% more people are needed than a simple random sample of equivalent precision would require, not because the formula is being cautious, but because that is genuinely how much statistical information is lost to clustering.

02The formula

DEFF = 1 + (m − 1)ρ m is the average number of people sampled per cluster. ρ (rho) is the intraclass correlation coefficient.

Two things drive DEFF up, sampling more people per cluster, m, and clusters being more internally homogeneous, ρ. DEFF equals 1 exactly when ρ equals 0, meaning clusters are not actually more alike internally than the population at large. In that case, clustering costs nothing, and cluster sampling behaves just like simple random sampling.

03What ρ actually measures

The intraclass correlation measures how much of the total variation in an outcome is between clusters versus within them. A ρ near 0 means clusters are essentially interchangeable, most variation is between individuals regardless of which village they belong to. A ρ near 1 means clusters are highly distinct, knowing the village reveals almost everything about the individual.

Typical ρIndicator type
0.01 to 0.03Individual attitudes, knowledge, and awareness questions
0.03 to 0.07Behavioural indicators, practice, uptake, usage
0.05 to 0.15Health and household-level indicators, immunisation, water access, nutrition

These ranges are starting points, not universal truths. Where possible, use ρ from a prior round of the same survey or a closely comparable study, rather than a generic benchmark.

How to Set This in AnalyZ Solutions
  1. Choose Cluster or multi-stage as the sampling approach. This makes the design effect settings available in the calculator you are using.
  2. Enter DEFF directly, or calculate it from ICC. Enter a study-specific DEFF if one is already known, or switch to calculating from ICC and enter an assumed ρ and average cluster size.
  3. Review the value shown live. Where DEFF is calculated from ICC, the resulting value is displayed as the inputs are adjusted.

Design effect defaults to 1 in every calculator across AnalyZ Solutions, meaning no clustering effect assumed, and is always editable, whether or not the chosen sampling approach is cluster-based.

VIDEO WALKTHROUGH PLACEHOLDER, 90 SECONDS

A short screen recording showing these steps in the AnalyZ Solutions interface can be embedded here.

04Worked example

A household water-access survey samples 15 households per village. Based on a prior round, ρ is estimated at 0.06 for this indicator.

Calculation
Average cluster size, m15
ICC, ρ0.06
1.84
DEFF = 1 + (15 − 1) × 0.06

Applied to a base simple random sample of 385, this DEFF pushes the requirement to roughly 709 households, a substantial jump that would be easy to miss if DEFF were left at its default of 1.

05The practical lever, fewer people per cluster

Because DEFF grows with m, people per cluster, but ρ is largely fixed by the indicator, the most effective way to reduce DEFF, and therefore the total required sample, is usually to select more clusters and fewer people within each one, rather than fewer clusters sampled more intensively. This is a genuine design trade-off against field logistics costs, but it is often underused.
Frequently Asked Questions
Does DEFF apply to stratified sampling too?
Not in the same way. Stratification generally improves precision, or at worst leaves it unchanged, because it guarantees representation rather than introducing within-group correlation. DEFF is specifically about clustering, grouping units and sampling whole groups together, not about splitting the population into strata.
What if I genuinely have no idea what ρ is for my indicator?
Enter DEFF directly instead, using 1.5 to 2.0 as a reasonable planning default for most community-level surveys, and note this as an assumption in the protocol. Revisit it after a pilot if precision estimates turn out to matter a great deal for the decision at hand.
Does a higher DEFF mean the study design is flawed?
No. It means the study design has a real, unavoidable statistical cost that a simple random sample would not have. Cluster sampling is often the only practical option when a full individual-level sampling frame does not exist. DEFF makes that trade-off visible and accounted for, rather than hidden.

Ready to apply this to your own sample size?

Every AnalyZ Solutions sample size calculator supports DEFF and ICC directly.

Try it out
Related Guides